CCTV Cameras

Choose by the pixel density you need on the target, not by the megapixel number alone. 2 MP is 1920×1080, 4 MP is usually 2560×1440 (some models use 2688×1520) and 8 MP is 3840×2160, and with the same lens the pixels per metre grow in proportion to the horizontal pixel count. With a 90° lens, for example, a face reaches the identification density of 250 px/m at about 3.8 m with 2 MP, 5.1 m with 4 MP and 7.7 m with 8 MP. As a starting point, 2 MP is enough for a single doorway or a small room, 4 MP suits most shops, offices and yards, and 8 MP pays off in wide open areas or when you want to zoom into recordings, at the cost of a higher bitrate and a little more noise at night.

Not exactly: 8 MP (3840×2160) has four times the pixels of 2 MP (1920×1080), but only twice as many across the width and twice as many in height, so at the same distance and field of view you get twice the pixels per metre, not four times. The extra pixels only help if the lens is sharp, the focus is correct, there is enough light and the bitrate is high enough to keep the detail. That is why 8 MP is worth choosing when you need the extra reach or the ability to crop into the picture, and not simply because the number is bigger.

On the same sensor, more pixels means smaller pixels: an 8 MP pixel has only a quarter of the area of a 2 MP pixel, so it collects less light and shows more noise in the dark. A larger sensor helps: a 1/1.8 inch sensor has about 1.9 times the area of a 1/2.8 inch one (the fraction is a legacy naming convention, not the real size). The lens aperture matters just as much, because the light reaching the sensor is inversely proportional to the square of the F-number: an F1.0 lens admits about 2.6 times more light than F1.6 and 4 times more than F2.0. For dark scenes, compare resolution, sensor size and F-number together instead of looking only at the megapixel count.

“Starlight” has no formal standard: manufacturers use it for cameras with a fast lens (often F1.6 or faster) and a low-noise sensor that still give a usable colour picture in very little light, usually 0.01 lux or less on the datasheet. The lux figure is measured under favourable conditions (widest aperture, maximum gain, slow shutter), so a picture at that light level is dark and noisy and moving objects blur. For comparison, a full moon gives about 0.05 to 0.3 lux and a moonless clear night about 0.002 lux. Compare datasheet values only when the test conditions are the same, and judge a camera on a live picture of your own scene.

An infrared (IR) camera switches to black-and-white at night and lights the scene with invisible IR LEDs, while a full-colour camera keeps a colour picture at night by combining a very sensitive sensor and a fast lens with a visible warm or white light. Colour helps you describe clothing and vehicles, and a visible light can also deter intruders, but it is not covert and it lights up the area. Dual-light models combine both and switch between IR and warm light automatically or by schedule. Depending on the model, KDT datasheets list a warm-light range of about 20 to 90 m and, on dual-light models, an IR range of about 40 to 80 m.

The IR range on a datasheet (for example 40 m or 80 m) is normally quoted for ideal conditions: an open scene and a well-reflecting target. IR light spreads out, so its intensity falls roughly with the square of the distance: a target at twice the distance receives only a quarter of the light. A wide lens spreads the same IR power over a larger area, dark clothes reflect little, and rain, fog or insects near the lens scatter light back and reduce contrast. Treat the datasheet figure as a best case, expect usable detail at clearly shorter distances, and add a separate IR or white-light illuminator for long ranges.

WDR (wide dynamic range) lets a camera show both the bright and the dark parts of a scene, such as a sunlit street seen from a dim lobby, without burning out the highlights or losing the shadows. True (sensor-level) WDR merges several exposures per frame and is specified in dB: 120 dB means a brightness ratio of about 1,000,000 to 1, while a single exposure covers roughly 60 dB (1000 to 1). Digital WDR (DWDR) only brightens the shadows of a single exposure by software, so it cannot recover detail the sensor has already clipped and it adds noise. True WDR can show ghosting on fast-moving objects, so use it for high-contrast entrances and windows, and check that the datasheet gives a dB figure and not only DWDR.

BLC (backlight compensation) raises the exposure for the subject when a strong light is behind it, for example a person in front of a bright window, and lets the bright background burn out. HLC (highlight compensation) darkens or masks very bright spots such as headlights and lamps so that the face or plate next to them stays visible, which helps at night at gates and car parks. DNR (digital noise reduction, 2D or 3D) reduces the grain that appears in low light, but a high 3D DNR level leaves trails behind moving objects. Use WDR first for high-contrast scenes, add BLC or HLC only for specific problems, and raise DNR gradually while watching a moving subject.

Many lamps powered from the 50 Hz mains, such as fluorescent and simple LED lamps, flicker at 100 Hz, and if the camera's exposure time does not match that, dark bands roll through the picture. Set the camera's power-line frequency or anti-flicker option to 50 Hz, the mains frequency in Iran, or use exposure times that are multiples of 1/100 s (1/100, 1/50 or 1/25 s). Avoid fixing a very short shutter such as 1/500 s indoors, because it makes the bands more visible.

The shorter the focal length, the wider the view and the fewer pixels per metre on the target. For a common 1/2.8 inch 16:9 sensor (about 5.6 mm wide), the scene width is roughly the distance × 5.6 ÷ the focal length in mm: at 10 m a 2.8 mm lens covers about 20 m (about 90°), a 4 mm lens about 14 m (70°), a 6 mm lens about 9.3 m (50°) and a 12 mm lens about 4.7 m (26°). Real lenses differ a little, especially wide ones below about 3 mm, and the same focal length gives a wider angle on a larger sensor, so use the angle on the datasheet or the Field-of-view calculator among KDT's after-sales services on the website. Choose the longest lens that still covers the whole area you need, because that gives the most pixels on the target.

A fixed lens has a single focal length (for example 2.8 or 4 mm); it is simple and usually the brightest, but you must choose the right one before installation. A varifocal lens covers a range (for example 2.8 to 12 mm) that you adjust by hand on the camera, and a motorized varifocal lens lets you set zoom and focus from a computer or phone without a ladder; many KDT motorized models, for example, cover 2.7 to 13.5 mm, with a horizontal angle of roughly 94° to 112° at the wide end and 34° to 36° at the long end. Choose varifocal or motorized when the exact distance is uncertain, the view may change later or the camera is hard to reach, and a fixed lens when the framing is known and the budget is tight.

Optical zoom changes the focal length of the lens, so a distant target gets more real pixels and more detail; a KDT 32× speed dome, for example, has a 4.8 to 154 mm lens. Digital zoom only crops and enlarges the image, so it adds no detail and the picture just becomes blockier. A high-resolution camera does let you crop into recordings with acceptable quality (an 8 MP image cropped to its central quarter still has about 2 MP), but that is no substitute for a longer lens, so compare cameras by their optical zoom.

DORI stands for Detection, Observation, Recognition and Identification, four image-quality levels that IEC 62676-4 defines by the pixel density on the target. Detection is 25 px/m (you can tell that a person or vehicle is present), Observation 62.5 px/m (you can see clothing and what people are doing), Recognition 125 px/m (you can tell whether it is someone you have seen before) and Identification 250 px/m (you can identify a person, roughly 40 pixels across a face 16 cm wide). The values assume good lighting, correct focus and a target that is not moving very fast or seen from a steep angle, so poor light, motion or heavy compression call for more pixels. Many datasheets list DORI distances calculated from these densities and the lens angle, which lets you compare cameras on the same basis.

Work out the pixel density at the target: the scene width at distance d is W = 2 × d × tan(H/2), where H is the horizontal angle of view, and the density is p = N/W, where N is the number of horizontal pixels. Rearranged, the longest distance for a given density is d = N/(2 × p × tan(H/2)). For example, a 4 MP camera (N = 2560) with a 90° lens sees a scene 20 m wide at 10 m, which is 128 px/m; it identifies (250 px/m) up to about 5.1 m, recognises (125 px/m) up to 10.2 m, observes (62.5 px/m) up to 20.5 m and detects (25 px/m) up to 51.2 m. Real distances are shorter in poor light, so keep a margin, or let the Field-of-view calculator on the KDT website do the work for a specific model.

At entrances 2.5 to 3 m is a common mounting height, aimed so that people walk towards the camera rather than across it. The downward angle to a face (assumed to be 1.6 m above the floor) is roughly atan((h − 1.6)/d), where h is the mounting height and d the horizontal distance in metres: from h = 3 m, a face 5 m away is seen from about 16° and one 2.4 m away from about 30°. Keep the angle small, aiming for about 15° or less and avoiding much more than 30°, because a steep view from above hides facial detail. Also avoid pointing the camera at the sun or a bright window, and mount it out of easy reach.

A bullet camera has a long body with a sun shield; it is easy to aim, visible as a deterrent and usually has the longest IR range, so it suits building walls, yards and long corridors. A dome camera hides the direction of its lens under a vandal-resistant cover, which suits ceilings and public places, but a dirty or scratched dome can reflect the IR light back into the lens. A turret camera has no dome cover, so the IR LEDs are much less likely to reflect inside the housing, and it can be adjusted on three axes, which makes installation quick. Choose by location and risk: bullet for long range and visible deterrence, dome for discreet or vandal-prone places, and turret for quick, glare-free installation.

A PTZ camera pans, tilts and zooms (KDT speed domes offer, for example, 32× and 38× optical zoom), so one camera can survey a large area such as a car park, a yard or a port and zoom in on details. It can look in only one direction at a time, so events outside its current view are missed unless it runs a patrol or auto-tracking, and its moving parts wear over time. For continuous recording of key points such as entrances use fixed cameras, and add a PTZ to follow and zoom in on what the fixed cameras show. Mount it on a rigid support, because vibration ruins a zoomed image, and plan for more power, since speed domes usually need 802.3at PoE (25 to 30 W) or more.

A fisheye camera sees up to 180° (a whole room when mounted on the ceiling), and a panoramic model with two lenses joins two views into one wide picture, so one camera can replace several in lobbies, shops and halls. The pixels are spread over a very large area, so at a distance the detail is much lower than with a fixed camera of the same resolution, and the picture is usually converted by software into normal views. Use it for an overview and for counting people, and pair it with a fixed camera where you need to identify faces, such as at entrances.

The IP code of IEC 60529 has two digits: the first (6) means the enclosure is dust-tight, and the second gives the water protection. IP66 means it withstands powerful water jets (12.5 mm nozzle, 100 litres per minute), IP67 means it survives temporary immersion (up to 1 m for 30 minutes) and IP68 means continuous immersion under conditions set by the manufacturer. Formally IP67 does not include the jet test, but for rain and normal outdoor use either rating is suitable. Many KDT outdoor cameras are rated IP67, while covert pinhole models are IP44 (splash-proof) and not intended for exposed outdoor use.

IK10 is a rating of IEC 62262 for the impact resistance of an enclosure: the housing withstands an impact energy of 20 joules, for example a 5 kg mass dropped from 40 cm. It is the highest rating usually seen on CCTV cameras and gives good protection against blows, kicks and thrown objects, and many KDT domes and bullet cameras are rated IK10. It does not make a camera indestructible, because repeated hits, a hammer or a crowbar can still damage it, and the rating refers to the housing, not to the bracket or the cable. In exposed places mount the camera out of easy reach and protect the cable.

The IP rating holds only if the connections are sealed too: use the supplied waterproof boot or cap on the RJ45 connector, or a weatherproof junction box, and leave a drip loop in the cable so that water runs off before it reaches the connector. Water can also travel inside the cable jacket, so never leave a cable end unsealed and do not point the cable entry upwards. Check the operating temperature range on the datasheet (typically −30 to +60 °C for KDT cameras) and the surge-protection level (2 kV or 6 kV on KDT models), and ground the camera and its pole, because built-in protection does not replace proper grounding. Mount the camera under an eave or with a sun shield where possible, which reduces heat, rain streaks and lens contamination.

The PoE standards define how much power a switch port can supply: 802.3af (PoE) up to 15.4 W, of which at least 12.95 W reaches the camera; 802.3at (PoE+) up to 30 W (25.5 W at the camera); and 802.3bt up to 60 W or 90 W (51 W or about 71 W at the camera). Most fixed KDT cameras list 802.3af with a maximum consumption of roughly 6 to 13 W, speed domes list 802.3at (25 to 30 W) and the explosion-proof models list 802.3bt (60 W). Add up the maximum consumption of all cameras and keep the total within the switch's PoE budget with some margin (20% is a common rule): 8 cameras at 9 W need 72 W, so a switch with a 120 W budget is enough. The cable run is limited to 100 m, and pure-copper cable (not copper-clad aluminium) is needed for full PoE power.

Check the power first: at night the IR LEDs (and on some models a heater) raise consumption, so a camera on a PoE port or switch that is already at its limit, or at the end of a long or poor cable, can reboot exactly then. Then check the cable: Ethernet is limited to 100 m, and copper-clad aluminium (CCA) cable and badly crimped or wet RJ45 connectors cause drops. Two devices with the same IP address, an overloaded switch uplink or a network loop also make cameras disappear, so give each camera a fixed or reserved address and test by connecting the camera directly to a laptop with a short, known-good cable. If the problem remains, update the firmware and look at the camera and switch logs for reboot or PoE events.

H.265 (HEVC) typically needs 30 to 50% less bitrate than H.264 at similar quality, which saves bandwidth and storage, but it takes more processing power to decode, and older recorders, players and some browsers may not support it. H.264+ and H.265+ are vendor-specific smart-codec modes that cut the bitrate further in scenes with little movement; they behave differently from brand to brand, so test them with your recorder. Use H.265 when your recorder and viewing devices support it, and keep H.264 for compatibility, for example on the sub-stream that phones or older PCs use. MJPEG creates very large streams and is only useful for snapshots or special software.

The main stream carries the full resolution and is normally the one you record; the sub-stream is a low-resolution copy (for example about 640×360 at well under 1 Mbps) used for live multi-camera views and phones, and a third stream can serve a mobile app or another system at the same time. Most KDT cameras list triple-stream output, meaning three independent streams. Watching a 16-camera grid on the sub-streams needs a fraction of the bandwidth and decoding power of 16 full-resolution streams, and you switch to the main stream only when you open one camera full screen. Set the main stream for your recording needs and the sub-stream low enough for your slowest viewing device or link.

ONVIF is an open standard that lets cameras, recorders and software from different manufacturers work together. Profile S covers the basics (live video streaming, device discovery and PTZ control), Profile G covers edge storage (searching and playing back recordings kept on the camera's memory card), and Profile T covers advanced streaming, including H.265, imaging settings and events such as motion or tampering. KDT datasheets list Profile S on entry-level cameras and some specialised models and Profiles S, G and T on most Smart-series models, so check that the camera and the recorder support the same profiles.

Only a product whose manufacturer has passed ONVIF's official conformance test and whose conformance is registered in ONVIF's public database is ONVIF conformant, and only such products may use the ONVIF profile logos; “compatible” is just the manufacturer's own statement with no public listing behind it. Conformance applies to a specific model, firmware version and profile, and is listed on the official ONVIF website, so check the exact model there or ask for the declaration of conformance. Even a conformant device may not expose every extra function (such as analytics settings or smart-codec modes) through ONVIF, so live view and recording usually work but advanced features may need the manufacturer's own protocol or SDK. Before buying for a mixed-brand system, compare the profile lists of the camera and the recorder and test one camera first.

Usually yes, through ONVIF or RTSP: live view, recording and PTZ control normally work when both sides support the same ONVIF profile, and KDT datasheets also list an API, a CGI interface and an SDK for integrators. Advanced functions such as analytics events, smart-codec modes, playback from the camera's memory card and some image settings may work fully only with KDT recorders and software, so check the other system's feature list first. Add one camera as a test and confirm the stream, the frame rate, H.265 support and the events you need before rolling out the whole project. If the other system asks for a stream address, use the RTSP address of the main or the sub-stream.

RTSP (Real Time Streaming Protocol) is the standard way to request a live video stream from a camera; the camera listens on port 554 by default and sends the video itself over RTP. In a player such as VLC you open a network stream with an address of the form rtsp://username:password@camera-IP:554/path, where the path (which selects the main or the sub-stream) is different for each manufacturer and is given in the camera's manual, or can be read from the camera through ONVIF. If the picture stutters over Wi-Fi or a VPN, switch the transport from UDP to TCP. RTSP is not encrypted by default, so use it only inside a protected network and never expose it to the internet.

Edge analytics run on the camera itself and raise an event or start recording when something crosses a virtual line, enters or leaves a zone, loiters or moves fast; many KDT models list line crossing, area intrusion and region entrance and exit, while people and vehicle classification is limited to selected models, so check the datasheet. Their limits come from the camera's small processor and the picture: a target that is too small, too far, partly hidden or seen at a steep angle is missed, and rain, snow, insects, shadows, headlights and swaying trees cause false alarms. Classifying people and vehicles is more reliable than plain motion detection, and as a rule of thumb the target should have at least the Observation density (62.5 px/m, about 100 pixels for a standing person) at the far edge of the zone. Treat analytics as an aid to operators, not a replacement for them.

Draw the detection zones only where events matter, excluding trees, flags, busy roads, reflective surfaces and the sky, and use a line or zone rule with a direction and a minimum size instead of plain pixel motion. Fix the picture first: a correct exposure, no strong IR reflection from nearby walls and a clean lens, because noise and glare at night are common triggers. Insects and spider webs near the lens or the IR LEDs create moving spots at night, so clean the camera regularly and keep it away from lamps that attract insects. Raise the sensitivity in steps, test at night and in bad weather for a week or two, and use a schedule or a minimum dwell time of a few seconds before an alarm is sent.

The camera records to its own microSD card, continuously or on an event, and KDT datasheets list card slots for 128 GB up to 512 GB depending on the model; on models that list ANR, the camera fills gaps in the recorder's footage after a network outage by sending the stored video afterwards. Capacity depends on the bitrate: at 4 Mbps a stream uses about 43 GB per day, so a 256 GB card holds roughly 6 days, and at 2 Mbps about 12 days. Treat the card as a backup and not as the main recording, because it is small, wears out and can be stolen together with the camera. For hard disk sizing, use the Storage calculator among KDT's after-sales services on the website.

Use a high-endurance (industrial or surveillance-grade) microSD card designed for continuous recording, because ordinary consumer cards are built for occasional writing and wear out much faster under 24/7 recording. Choose a capacity within the maximum on the camera's datasheet (for example 256 GB) and a speed class of at least Class 10 (U1), buy from a reliable seller because counterfeit cards are common, and format the card in the camera before use. Check the card status in the camera's web page from time to time, replace it before its rated endurance is used up and keep a spare on site.

Start with the basics: change the factory password at the first login to a long, unique one (at least 12 characters), remove or disable unused accounts, and never leave a camera on default credentials, because botnets scan the internet for exactly that. Keep the firmware up to date and turn off the services you do not use, such as UPnP, P2P or cloud access, Telnet, SSH and FTP, and use HTTPS for the web interface. Put the cameras on a separate network segment (VLAN) that only the recorder and the administrator's computer can reach, and do not give them direct internet access. Review the user accounts and access rules once or twice a year.

Port forwarding exposes the camera's web page and RTSP port to the whole internet, where automated scanners find them quickly and try default or weak passwords and known vulnerabilities. UPnP lets a device open such ports on the router by itself without asking you, and P2P (cloud) access lets an app reach the camera through a server using its serial number; both are convenient but widen the attack surface, so turn them off if you do not need them. For remote viewing, use a VPN into the site (or a secure remote-access service you trust) and keep the cameras themselves unreachable from outside. If you must publish something, forward only the one port needed, use HTTPS and a strong password.

Create a separate VLAN for the cameras (for example VLAN 20), assign the camera ports on the switch to it, and bring it to the recorder either by connecting the recorder to the same VLAN or over a tagged trunk. On the router or firewall allow only the recorder or video management software and the administrator's computer to reach the cameras, and block the cameras from reaching the internet and other VLANs. Automatic camera discovery (ONVIF WS-Discovery) normally works only inside one VLAN, so add cameras by IP address when the recorder is in another one. Use fixed or reserved addresses and disable unused switch ports; the exact VLAN steps depend on the switch model.

Check for updates when you install the camera and then every 6 to 12 months, or whenever a security fix is announced, because firmware updates also correct bugs and improve compatibility. Download the file for the exact model from the download section of the KDT website, back up the settings, and never interrupt the power or the network during the update. Update one camera first and check that the recorder still works with it, then update the rest in batches, and keep the previous version file in case you need to go back.

The usual cause is IR light bouncing back into the lens: a wall, pole or window close in front of the camera, or a dirty, wet or scratched dome, or insects and spider webs on the lens or the LEDs. Clean the dome and lens, move the camera or aim it away from nearby surfaces, and never place an IR camera behind window glass (turn its IR off and light the area instead). If the picture is sharp by day but soft at night, the focus has shifted with the IR light, so refocus at night or use a motorized lens with auto focus; long exposures at night also blur moving people, so limit the maximum shutter time. Turret cameras (no dome cover) and cameras with a separate illuminator are much less prone to reflections from a cover.

Yes. Cameras and a recorder (or the camera's own microSD card) work entirely on the local network, so recording and live viewing on site need no internet. The internet is needed only for remote viewing from outside, cloud services, push notifications and online updates; use a local NTP server (for example on the router or recorder) so that timestamps stay correct. For remote access without exposing the cameras, use a VPN rather than port forwarding.